By Hadas Gold, CNN
(CNN) — Elon Musk’s AI chatbot, Grok, has been flooded with sexual images of mainly women, many of them real people. Users have prompted the chatbot to to “digitally undress” those people and sometimes place them in suggestive poses.
In several cases last week, some appeared to be images of minors, leading to the creation of images that many users are calling child pornography.
The AI-generated images highlight the dangers of AI and social media – especially in combination – without sufficient guardrails to protect some of society’s most vulnerable. The images could violate domestic and international laws and place many people, including children, in harm’s way.
Musk and xAI have said that they are taking action “against illegal content on X, including Child Sexual Abuse Material (CSAM), by removing it, permanently suspending accounts, and working with local governments and law enforcement as necessary.” But Grok’s responses to user requests are still flooded with images sexualizing women.
Publicly, Musk has long advocated against “woke” AI models and against what he calls censorship. Internally at xAI, Musk has pushed back against guardrails for Grok, one source with knowledge of the situation at xAI told CNN. Meanwhile, his xAI’s safety team, already small compared to its competitors, lost several staffers in the weeks leading up to the explosion of “digital undressing.”
‘Digitally undressing’ people
Grok has always been an outlier compared to other mainstream AI models by allowing, and in some cases promoting, sexually explicit content and companion avatars.
And in contrast to competitors such as Google’s Gemini or OpenAI’s ChatGPT, Grok is built into one of the most popular social media platforms, X. While users can talk to Grok privately, they can also tag Grok in a post with a request, and Grok will respond publicly.
The recent surge in widespread, non-consensual “digital undressing” began in late December, when many users discovered they could tag Grok and ask it to edit images from an X post or thread.
Initially many posts requested Grok put people in bikinis. Musk reposted images of himself and others, like longtime nemesis Bill Gates, in bikinis.
Researchers at Copyleaks, an AI detection and content governance platform, found that the trend may have started when adult-content creators prompted Grok to generate sexualized imagery of themselves as a form of marketing. But almost immediately “users began issuing similar prompts about women who had never appeared to consent to them,” Copyleaks found.
Researchers at AI Forensics, a European non-profit that investigates algorithms, analyzed over 20,000 random images generated by Grok and 50,000 user requests between December 25 and January 1.
The researchers found “a high prevalence of terms including ‘her’ ‘put’/’remove,’ ‘bikini,’ and ‘clothing.’” More than half of the images generated of people, or 53%, “contained individuals in minimal attire such as underwear or bikinis, of which 81% were individuals presenting as women,” the researchers found. Notably, 2% of images depicted people appearing to be 18 years old or younger, the researchers found.
AI Forensics also found that in some cases, users requested minors be put in erotic positions and that sexual fluids be depicted on their bodies. Grok complied with those requests, according to AI Forensics.
Although X allows pornographic content, xAI’s own “acceptable use policy” prohibits “Depicting likenesses of persons in a pornographic manner” and “The sexualization or exploitation of children.” X has suspended some accounts for these kinds of requests and removed the images.
On January